At this point in the research project, i.e. in the initial data collection phase before the students start their new training, we cannot say anything about the effect of entrepreneurship education. We can analyze differences between groups in our study group, and give an initial picture of how our investigation will evolve and how en- trepreneurship education can influence these groups in different ways. We will focus on our five entrepreneurial self-efficacy measurements, and how these relate to entrepreneurial attitudes, intentions, and entrepreneurial behaviour.
Before the first round of data collection was carried out on a large scale, we made many pilot studies, which included a total of approx. 400 students in different courses, all of which gave results that supported the validity of our scale and convinced us that we could continue with the ”real” investigation. Demographics and background variables for our study group are presented in Table 4.2 below.
2011
40 Study of Entrepreneurship Courses at the University Graduate Level
Total number of respondents 556
- Respondents screened out because of lacking answers 149
Number of respondents included in the analysis 407
Descriptive statistics of the 495 respondents Gender
- Men 61.7%
- Women 38.3%
Age (mean) 25,0
Exchange students 30.5%
Have close family members (parents, siblings, uncles/aunts)who are self-employed 58.5%
Have taken a course or training program in the past 32.2%
Have participated in extra-curricular entrepreneurial activities in the past 2%
Years of part-time work experience (mean) 6.4 years
Years of full-time work experience (mean) 2.3 years
Years of full-time education (incl. elementary, secondary and tertiary level) 15.8 years Have, alone or together with others, started a business in the past 22.4%
Does, alone or together with others, operate a business today 13%
Table 4.2 - Basic characteristics of sample and data
4.6.1. The Reliability and Validity of the Measurements
When a new design of a questionnaire is used, it is important to validate whether it actually measures what we want to measure. We used a statistical test called Cronbach’s Alpha test to validate that the various sub-questions in the questionnaire was explained by the constructs, i.e. the five different self-efficacy constructs, which we had derived from the previous scales. In order to make sure that there is an internal connection, i.e. that the questions are measurements of the same as the other related issues in the design, the Cronbach Alpha score should be higher than 0.70. As indicated in Table 4.3 below, our scores are well above 0.70. In fact, our scores are above 0.80 for all constructs except for the planning construct, which indicates a good correlation.
Construct Cronbach’s Alpha All Cronbach’s Alpha Experiment Cronbach’s Alpha Control Creativity 0.84 0.84 0.81 Planning 0.75 0.77 0.74 Marshalling 0.81 0.81 0.80 Manage Uncertainty 0.84 0.86 0.81 Financial Knowledge 0.85 0.88 0.81 Entrepreneurial Attitudes 0.84 0.84 0.81 Entrepreneurial Intentions 0.88 0.84 0.82
Table 4.3 - Cronbach‘s Alpha scores. The internal consistency of our constructs.
We also wanted to test whether our scale was reliable in comparison to students in the control group, could the students understand the questions? We then performed separate Cronbach’s Alpha tests of the two groups, which are also presented in Table 3.
It is clear that the scores for the control group are lower, but not much, and they are all well above the critical 0.70 score, which indicates a reliable internal consistency and that the students in the control group understand the wording of the questions.
4.6.2. Differences between Students in the Experiment Group and Control Group
To make an adequate impact analysis, the subjects in the experiment group (i.e. those receiving treatment) and control group (i.e. those not receiving any treatment) should be very similar. We therefore undertook statisti- cal tests, called t-tests to examine how the mean values for the two groups were different on each design and whether the differences between the two groups was statistically significant. The results are presented in Figure 4.2.
Figure 4.2 - The mean values between the experiment group and control group in our five entrepreneurial self-efficacy constructs, attitudes and intentions.
0 10 20 30 40 70 50 80 60 90 100
Creativity Planning Marshalling Manage
Uncertainty Knowledge Financial Attitudes Intentions
Experiment group (n=108) Control group (n=148)
2011
42 Study of Entrepreneurship Courses at the University Graduate Level
It should be noted that the various constructs are measured with a different number of questions, but all use a scale of 1-7. For clarity, we have converted these constructs to the same unit, so they can be compared with each other. All statistical tests are performed with the original data.
There was a significant difference between the two groups on all structures on a 1 per cent level, except planning and marshalling which only showed a significant difference at a 5 per cent level, which is the significance level we use26.1This indicates that students in the experiment group and control group are different from each other
in terms of entrepreneurial self-efficacy, attitudes and intentions. It is very likely that some of the students on the courses are already very positive about entrepreneurship even before they start the course. This is not ideal when you want to perform an impact analysis, but as long as the variables are known, they are simple to check for. There may be other factors besides the choice of training, which may explain the difference between the two groups of students, such as whether they are nascent entrepreneurs or not, which we will return to below. 4.6.3. Nascent Entrepreneurs
Nascent entrepreneurs, i.e. individuals who are actively trying to start a new business, are rare in the population and are often quite difficult to identify, in contrast to e.g. small business owners, employees and unemployed. In our study, we identify nascent entrepreneurs by asking this question: Are you trying to start a business for real as opposed to just evaluating an idea out of interest or as part of an academic exercise? If they responded positively to this question, they were asked to tick off a list of 19 entrepreneurial activities. If they selected two or more, we identified them as nascent entrepreneurs. 88 individuals, representing 22 percent, were identified as nascent en- trepreneurs. 67 of these nascent entrepreneurs were found in the experiment group and 21 in the control group. If our scale measures the actual entrepreneurial self-efficacy as well as attitudes and intentions, there should be a large and significant difference between ”regular” students and students who are nascent entrepreneurs. We have therefore conducted statistical tests of the mean difference (t-test) between the two groups. The results are presented in Figure 4.3 below.
26. The level of significance tells how confident we are that our results are correct. We are 99 per cent sure when the level is 1 per cent.
Figure 4.3 - The average differences between ordinary students and nascent entrepreneurs in our five entrepreneurial self-effica- cy constructs, attitudes and intentions.
0 10 20 30 40 70 50 80 60 90 100 Nascent entrepreneurs (n=88) Ordinary students (n=318)
Sig. Sig. Sig. Sig. Sig. Sig. Sig. Creativity Planning Marshalling Manage
Figure 4.4. - The mean difference between the experiment group and the control group in the five entrepreneurial self-efficacy constructs as well as Attitudes and Intentions when controlled for entrepreneurial experience. (Sig. stands for significant diffe- rence.)
The difference between the two groups evens out when we control for entrepreneurial experience, and the statisti- cal difference between the two groups actually disappears for Planning, Marshalling, Managing Uncertainty, and Financial Knowledge. The Creativity construct as well as the Attitudes and Intentions constructs remain significan- tly different. This suggests that the difference between the ”regular” students, i.e. those without entrepreneurial experience in the two groups not is so significant, especially when it comes to entrepreneurial self-efficacy. This is of great interest to us because this group of students most likely is influenced by entrepreneurship education. If a student already has entrepreneurial experience, education can improve individual knowledge and skills in the field, but it will be difficult to argue that it is because of the entrepreneurship education that the person has chosen to pursue a career as an entrepreneur.
It is evident that the differences between the two groups are large. The difference for these groups is significant at a 1 percent level for all constructs except Programming, which is significant at the 5 percent level. In other words, our measurements seem to measure entrepreneurial self-efficacy, attitudes and intentions in an appro- priate manner. The most interesting aspect of the bars in Figure 4.3 is, however, when they are compared to the bars in Figure 4.2. The values for the nascent entrepreneurs are much higher than for the experiment group; the values for the ”regular” students, i.e. students who are not currently trying to start a business in Figure 4.3., are almost identical with the students in the control group in Figure 4.2.; despite the fact that the group also includes 170 students from the experiment group. This indicates that the difference between the two groups may be driven by entrepreneurial experience. To test whether the differences between ”ordinary” students in the two groups disappear if we control for entrepreneurial experience, we conducted a test in which students who are nascent entrepreneurs, run a business currently, or have run a business, were not included. In Figure 4.4., the results of this test are presented.
0 10 20 30 40 70 50 80 60 90 100 Experiment group (n=108) Control group (n=148)
Sig. Sig. Sig.
Creativity Planning Marshalling Manage
2011
44 Study of Entrepreneurship Courses at the University Graduate Level
4.6.4. The Gender Distribution
There are 156 women in our sample (experiment and control groups), corresponding to approx. 38 per cent. They are fairly uniformly distributed with 80 in the experiment group and 76 in the control group, so that the gender effect is marginal. Women are on average far less likely to pursue a career as an entrepreneur27,1which also is
evident in our data. Of the 112 nascent entrepreneurs in our total sample, only 30 (27%) are women. When we test for differences between men and women for the seven constructs, the results are clear. As shown in Figure 4.5., women have significantly lower values on all constructs except Planning and Entrepreneurial Attitudes.
27. Among adult entrepreneurs 73 percent are men and 27 percent are women (in 2007). Source: The Danish Enterprise and Con- struction Authority. www.ebst.dk
Figure 4.5. – The mean difference between men and women in our five entrepreneurial self-efficacy constructs and Attitudes and Intentions. (Sig. stands for significant difference.)
These differences may be superficial, if the case is that women are more truthful about their own abilities. The difference between men and women with regard to nascent entrepreneurs, however, indicates that these are the true values and that women on average have significantly lower entrepreneurial self-efficacy in relation to all skills, except Planning. There is a real gender effect in our sample. An interesting observation is that we cannot find a significant difference between men and women in terms of attitudes to entrepreneurship. Based on these results, it should be seen as positive that there is such a significant amount of women (40%) in our experiment group.